Escherichia coli Topoisomerase IV E Subunit and an Inhibitor Binding Mode Revealed by NMR Spectroscopy*

Bacterial topoisomerases are attractive antibacterial drug targets because of their importance in bacterial growth and low homology with other human topoisomerases. Structure-based drug design has been a proven approach of efficiently developing new antibiotics against these targets. Past studies have focused on developing lead compounds against the ATP binding pockets of both DNA gyrase and topoisomerase IV. A detailed understanding of the interactions between ligand and target in a solution state will provide valuable information for further developing drugs against topoisomerase IV targets. Here we describe a detailed characterization of a known potent inhibitor containing a 9H-pyrimido[4,5-b]indole scaffold against the N-terminal domain of the topoisomerase IV E subunit from Escherichia coli (eParE). Using a series of biophysical and biochemical experiments, it has been demonstrated that this inhibitor forms a tight complex with eParE. NMR studies revealed the exact protein residues responsible for inhibitor binding. Through comparative studies of two inhibitors of markedly varied potencies, it is hypothesized that gaining molecular interactions with residues in the α4 and residues close to the loop of β1-α2 and residues in the loop of β3-β4 might improve the inhibitor potency.

Bacterial topoisomerases are attractive antibacterial drug targets because of their importance in bacterial growth and low homology with other human topoisomerases. Structure-based drug design has been a proven approach of efficiently developing new antibiotics against these targets. Past studies have focused on developing lead compounds against the ATP binding pockets of both DNA gyrase and topoisomerase IV. A detailed understanding of the interactions between ligand and target in a solution state will provide valuable information for further developing drugs against topoisomerase IV targets. Here we describe a detailed characterization of a known potent inhibitor containing a 9H-pyrimido [

4,5-b]indole scaffold against the N-terminal domain of the topoisomerase IV E subunit from Escherichia coli (eParE). Using a series of biophysical and biochemical experiments, it has been demonstrated that this inhibitor forms a tight complex with eParE. NMR studies revealed the exact protein residues responsible for inhibitor binding.
Through comparative studies of two inhibitors of markedly varied potencies, it is hypothesized that gaining molecular interactions with residues in the ␣4 and residues close to the loop of ␤1-␣2 and residues in the loop of ␤3-␤4 might improve the inhibitor potency.
The emergence of antibiotic resistant bacteria is a growing and serious health problem (1). Despite the development of a limited number of new antibacterial drugs (2), there is still a shortfall of available antibiotics to effectively tackle the issue of drug-resistant bacteria (3). It is therefore important to understand the mechanisms of drug interactions on a molecular level to further our development of new antibacterial agents. Bacterial topoisomerases are well studied drug targets that are present in all bacteria and are essential for bacterial growth (4). These enzymes exhibit low structural homology with human topoisomerases and are therefore attractive drug targets. Both the DNA gyrase B subunit (GyrB) 2 and DNA topoisomerase IV E subunit (ParE) contain ATP binding pockets and are respon-sible for DNA replication (5,6). Small molecule inhibition of this pocket is plausible, and a number of lead compounds have been developed targeting this pocket (7).
The availability of x-ray crystal structures of both ParE and GyrB have enabled structure-guided drug discovery (8). Past work done in the group has focused on using fragment-based approaches to develop inhibitors against GyrB (9). Further fragment-based approaches have also developed pyridylurea chemical leads against GyrB/ParE from both Gram-positive and Gram-negative bacteria (10). Gratifyingly, it has been shown that another class of inhibitors with a 9H-pyrimido [4,5-b]indole scaffold exhibited antibacterial activities (11).
X-ray crystal structures of GyrB/ParE-inhibitor complexes provide valuable information for understanding their interactions and, in turn, aid rational drug design (12)(13)(14). The ATP binding site is located at the N-terminal domain of GyrB/ParE, which makes it feasible to use a construct with a molecular weight of ϳ24 kDa to study target and drug interactions using different biophysical methods in vitro. Additionally, structural and biochemical studies revealed certain residues in the loop between ␤2 and ␣3 to be critical for the binding of novobiocin. Single mutations of residues in GyrB and ParE affected novobiocin potency by ϳ20-fold (15). Also, extensive work has been done in characterizing GyrB/ParE interaction with a bis-pyridylurea inhibitor (Fig. 1, inhibitor 1) using protein NMR spectroscopy (16 -18). It has been found that this compound binds to solution forms of GyrBs/ParEs in similar modes. However, the slight difference in the amino acid sequence of the two enzymes led to markedly different inhibition potencies (18).
In an effort to understand the binding of a new class of compounds against GyrB and ParE, a potent 9H-pyrimido [4,5-b]indole was chosen to be studied (11). This new class of inhibitors was shown to have broad-spectrum antibacterial activities against both Gram-positive and Gram-negative bacteria by targeting GyrB/ParE. Residues from the loop between ␤2 and ␣3 are found to be critical for inhibitor binding and act as a cover region to the binding pocket, which may affect the rational design of potent inhibitors (18). The x-ray structures of the GyrB/ParEs and the inhibitor with a 9H-pyrimido [4,5-b]indole scaffold demonstrated that the inhibitor has a molecular interaction with residues away from the cover region, which suggested that it may bind to the target with a different mode (11). Such a binding mode may contribute to its antibacterial activity against the growth of both Gram-positive and Gram-negative bacteria. It may be a novel class of potent antibacterial agents. A further binding study in solution would allow for comparison of x-ray crystal structures with the solution structure of the ligand-protein complex. Similar to previous work on bis-pyridylurea (16 -18), the binding characterization of inhibitor 2 ( Fig. 1) will provide useful information for further optimization and development of this series of compounds (19 -21). As compounds 1 and 2 have different structures and antibacterial activities, comparison of the binding modes of 1 and 2 with ParE will also provide insight into the antibacterial activity of 2, which will be helpful for developing potent ParE/GyrB inhibitors.
In this study, we used NMR and other biophysical methods to understand the interactions between the N-terminal domain of Escherichia coli ParE (eParE) and inhibitor 2 with a 9H-pyrimido [4,5-b]indole scaffold. First, we showed that this compound has an IC 50 of 9 nM against ParE from E. coli. Next, surface plasmon resonance (SPR) experiments corroborated the tight binding of 2 against eParE with a K D of 1.1 nM. Third, we identified residues that are critical for the binding of 2 based on chemical shift perturbation (CSP) in the presence of the inhibitor. Furthermore, a hydrogen-deuterium (H-D) exchange experiment revealed residues affected by the binding of 2. Last, we found that 2 can compete with 1 using 19 F NMR.

Results
Inhibitor 2 Binds to eParE and Inhibits Its Activity-GyrBs/ ParEs are validated drug targets for developing antibiotics. A pyrimido-indole scaffold series developed by Truis Therapeutics was shown to have a broad-spectrum activity against both Gram-positive and Gram-negative bacteria (11). Inhibitor 2 ( Fig. 1) was shown to be active against ParE of E. coli, with an IC 50 of 9 nM ( Fig. 2A). Further SPR study demonstrated that the dissociation constant (K D ) between eParE and 2 was 1.14 nM (Fig. 2B), which is more potent than inhibitor 1, whose K D was 902 nM against eParE (17).
Inhibitor 2 Can Stabilize eParE-In addition to SPR, thermal shift assays were also used to characterize ligand binding against eParE (9). As shown in Fig. 2C, in the absence of ligand, eParE exhibited a T m of ϳ48°C. With an excess amount of compound 2, eParE was found to be stabilized by more than 14°C, suggesting tight binding of the ligand to the protein. Two melting transition states could be observed with varying ligand-toprotein ratios (Fig. 2C). The curve corresponding to free eParE disappeared when the concentration of 2 was increased, which arose from the fact that the tight binding of the inhibitor separates the protein sample to both free and ligand-bound forms, which exhibited different melting temperatures.
NMR Study of eParE-2 Interactions-1 H, 15 N heteronuclear single quantum coherence (HSQC) spectra of eParE in the absence and presence of 2 were collected and compared (Fig.  3A). Clear chemical shift perturbations were observed, suggesting their interaction in solution. NMR titration experiments were conducted to qualitatively assess binding of 2 against eParE. As shown in Fig. 2B, two forms of peaks were observed in the presence of 2. The increase in peak intensity of residues in bound form and decrease in peak intensity of residues in free form were observed when 2 was titrated into a uniformly 15 N-labeled eParE (Fig. 3B). This clearly demonstrated slow exchange interactions in solution, supporting evidence of tight binding, similar to results obtained from SPR and thermal shift experiments.
Solution Structure of the eParE-2 Complex-The backbone resonance assignments of the eParE-2 complex were obtained by referring to the assignment of free eParE and analyzing three-dimensional NMR experiments (Fig. 4A). Most resonances in the 1 H, 15 N HSQC spectrum were assigned (Fig. 4A). The secondary structure of the complex was obtained based on the assigned backbone chemical shifts using TALOSϩ. Eight strands and five helices were identified in the eParE-2 complex, which is very similar to the eParE-1 complex (17). The overall secondary structural elements are similar to those obtained from the x-ray crystal structure (PDB code 1S14), except for the N-terminal 30 residues. The difference can be attributed to the flexibility of the region. More importantly, there was no obvi-  ous secondary structural change recorded in the protein when compound 2 was added. The only noticeable difference was a slightly shorter ␤1 when 2 was bound to eParE (Fig. 5). Overall, the results were similar to past studies of the eParE-1 complex, where overall secondary structures of eParE remain intact. Taken together, the results suggest that the ATP binding pocket of eParE are rigid and independent of the binding potencies of ligand toward it.
The chemical shifts of C␣ are sensitive to the secondary structures (22). TALOSϩ analysis suggested that 2 might not cause eParE to induce significant secondary structural changes. We still compared the C␣ and C␤ chemical shifts of the eParE-2 complex with those of free eParE (Fig. 5, A and B). Several residues exhibited changes in C␣ upon 2 binding (Fig. 5C) Inhibitor 2 Binds to the ATP Binding Pocket-The crystal structure of the E. coli GyrB in complex with an inhibitor (C3) with a 9H-pyrimido [4,5-b]indole scaffold has been solved (11), in which the inhibitor bind to the ATP binding pocket (Fig. 6A). The ATP binding domain of E. coli ParE has a similar structure as GyrB (Fig. 6B). We then conducted HADDOCK using ambiguous interaction restraints derived from the CSP results.
We could obtain a model showing inhibitor 2 binding to the ATP pocket (Fig. 6B). Residues that exhibited C␣ chemical shift changes were close to the inhibitor (Fig. 6C). To further confirm their binding mode in solution, we collected NOESY experiments for both free eParE and its complex (Fig. 6D). Two unambiguous NOEs between methyl protons of C1 of 2 and amide protons of Asp-69 and Val-165 were observed (Fig. 6D). It was not surprising to observe only two NOEs because most of the backbone amide protons are not proximal to 2 (within 5 Å). A water signal was observed for these two residues in all the NOESY experiments, suggesting that these backbone amide protons are exposed to the solvent, which is also consistent with the structure. Nevertheless, the NOE experiments confirm the orientation of the inhibitor in the eParE ATP binding pocket, which is similar to the x-ray structure of the GyrB-C3 complex (11).
Relaxation Analysis and H-D Exchange Experiment for eParE and Its Complex-To understand the dynamic characteristics of eParE in complex with 2, 15 N T 1 , T 2 , and NOE values were obtained (Fig. 7A). Consistent with the x-ray structure, most residues of eParE were stable in solution, which is characterized by high T 1 and NOE and low T 2 values (Fig. 7A). The C-terminal residues and N-terminal 20 residues are flexible, as evidenced by the low T 1 and NOE and high T 2 values. The loop (residues 93-115) between ␣3 and ␣4 is highly dynamic in solution. This loop exhibited no obvious change in the absence and presence of 2, which is consistent with the fact that it is not involved in the binding. Some residues from regions of residues 40 -50, 120 -140, and 180 -210 exhibited slightly higher het-NOEs in complex with 2, suggesting that compound 2 can stabilize eParE. It has been noted that free eParE is stable in solution, which is similar to its complex with 2 (Fig. 7A).
The H-D exchange experiment was conducted to further understand the effect of 2 on the structure and dynamics of eParE. In the absence of 2, quite a few residues from eParE were protected from exchanges, which is common for eParE with ␤-barrel structures (Fig. 7B). These residues were forming hydrogen bonds with others. A difference was observed for eParE in the absence and presence of 2. Residues in the ATP binding pocket, including Leu-35, Gln-37, Val-39, Asn-42, Val-44, Asp-45, Ala-52, Gly-71, Gly-73, Met-74, and Ser-164, were protected from exchanges, suggesting that these residues are critical for 2 binding (Fig. 7C). Inhibitor 2 may prevent deuterium from entering the ATP binding pocket by forming a tight complex with eParE. Several residues away from the ATP binding pocket were also protected from exchanges in the complex (Fig. 7B). These residues include Ile-58, Asn-120, Ser-123, Asn-143, Lys-146, Gly-162, Phe-174, and Arg-178. These residues may form backbone hydrogen bonds with nearby residues by changing their conformations slightly when eParE binds to 2.
On the other hand, amide protons of residues including Leu-59, Gln-63, Leu-122, Ala-190, Ala-192, Lys-203, Ile-206, and Thr-209 were protected in free eParE and exposed to solvent in the complex, suggesting that these residues may lose hydrogen bond formation with nearby residues in the complex. 19 F NMR of Protein-Inhibitor Interactions-As 2 contains one F atom, it will give only one resonance peak in the 19 F spectrum. The free 2 exhibited a resonance at Ϫ75 ppm (Fig.  8A), which is about 12 ppm difference from 1 that was used in a previous study (Fig. 8B) (17). Such a difference in chemical shift arises from the different chemical groups attached to the F atoms in the compounds (Fig. 1) because the chemical shifts of F atoms are sensitive to the chemical environment (23). In the presence of eParE, chemical shift changes were observed for 2 (Fig. 8). The presence of only one 19 F NMR resonance observed for the complex strongly suggest a single binding mode of 2 against eParE (20). Next, 19 F NMR experiments were conducted to demonstrate competition of the more potent inhibitor 2 against compound 1. A mixture containing eParE-1 was first prepared, and 19 F NMR was collected (Fig. 8B). One resonance peak of 1 was observed for the complex. In the presence of 2, two resonance peaks of 1 were observed (Fig. 8B), corresponding to the free and the eParE-bound 1, respectively (17). The signal of 2 was also observed at ϳϪ75.3 ppm (Fig. 8C). This set of established 19 F NMR experiments can be applied to further studies of ranking the potencies of different inhibitors against ParE using 1 as a reference compound.

Discussion
Structure-based drug design is a powerful technique that has been used in the development of potent antibacterial inhibitors (7,24,25). Both GyrB and ParE are validated drug targets and have attracted much drug discovery efforts around them (26,27). However, despite the advent of many potent inhibitors against the targets, no inhibitors against GyrB and ParE have been marketed (11). The availability of high-resolution x-ray structures of GyrB/ParE-inhibitor complexes has undoubtedly AUGUST 19, 2016 • VOLUME 291 • NUMBER 34

JOURNAL OF BIOLOGICAL CHEMISTRY 17747
been a boon toward the development of potent inhibitors against the target. In addition to x-ray structures, which provide structural information of a complex in its lowest energy state, NMR spectroscopy can provide information about target-inhibitor interactions in a more dynamic context. These solutionstate structural studies will certainly complement the understanding of strategies to improve drug potencies.
In previous studies, we studied interactions of GyrB/ParE and 1, which has a K D of 902 nM against eParE (17). We demonstrated that 1 binds to GyrBs/ParEs from both Gram-positive and Gram-negative bacteria in the same mode (18). In this study, we studied the interaction between eParE and 2 with a K D of 1.14 nM (Fig. 2) (11). We demonstrated that 2 formed a tight complex with eParE based on an enzymatic assay, SPR study, and thermal shift assay (Fig. 2). Further NMR study showed that residues in the ATP binding pocket were affected by the binding of 2. CSP induced by inhibitor binding could be used to guide docking studies using HADDOCK to obtain a model to understand the interactions (Fig. 6). For both 1 and 2, NOESY experiments can also be conducted to obtain distance restraints to validate and confirm the binding mode and determine complex structures (17) (Fig. 6). Our studies suggested that CSP-guided docking using HADDOCK is a useful and fast method to understand protein-drug interactions in a drug discovery process. It has been noted that the complex structure can also be solved for such a complex using NOE restraints because the orientation of the compound can be confirmed by using a F1-13 C/ 15 N-filtered F2-15 N-edited NOESY or 13 C-edited NOESY experiment in which the assignment of the 1 H, 15 N( 13 C) HSQC spectrum is available.
As 2 binds to eParE with a binding affinity (K D ϭ 1.14 nM) of ϳ900 times higher than that of 1 (K D ϭ 902 nM), the different residues of eParE affected by their binding may provide useful information to explain the difference in K D values. The CSP result suggested that both inhibitors affected the chemical environment of residues from ␣2, ␣3, ␣4, ␤2, ␤6, and the loop region between ␤2 and ␣3 (Fig. 4). Amide and amine protons of residues His-51, Arg-132, and Asp-133 of eParE exhibited significant chemical shift changes upon binding to 2 but were not affected significantly by 1. Residues in the loop between ␤1 and ␣2 and ␣4 of eParE exhibited obvious 13 C␣ chemical shift changes upon binding to 2, suggesting that gaining molecular interactions with residues close to these regions improved inhibitor potency. Based on the crystal structure of the  GyrB-C3 (an analog of 2) complex, there are salt bridges between Glu-50 and Arg-76 (corresponding to residues Glu-46 and Arg-72 in eParE) that can favor a -cation interaction with the aromatic ring of the inhibitor (Fig. 9A) (11). The side chains of His-55 (His-51 in eParE) form a close contact with those of Glu-50 and Arg-76. As the -cation interaction is critical for inhibitor binding, significant CSPs were observed for residues including Glu-46 and Arg-76 of eParE. In addition, residue His-51 was affected significantly, suggesting that side chain of His-51 might be important for stabilizing the complex by forming interactions with the side chains of Glu-46 and Arg-72 (Fig.  9B). Residues Arg-132 and Asp-133 showed significant chemical shift changes upon inhibitor binding (Fig. 9C). These two residues are far away from the ATP binding pocket. The side chain of Arg-132 can form a hydrogen bond with the backbone amide of Arg-72, which might be important for stabilizing the -cation interaction. The side chain of Arg-132 was proposed to have a hydrogen bond with the inhibitor through a water molecule (11). Arg-132 exhibited significant CSP upon inhibitor treatment (Fig. 4), suggesting that it might have a direct interaction with inhibitor 2. NOE experiments showed that the amide proton of Met-74 exhibited stronger NOEs with water protons upon ligand binding (Fig. 9D), suggesting that there are water molecules close to this residue that can favor the hydrogen bond formation as proposed (11). The loop between ␤2 and ␣3 containing residues Arg-72 to Met-74 was considered as a cover region to the binding pocket. Inhibitor 2 obtained its potency by forming interactions with both this cover region and residues in ␤3. ParE/GyrB N-terminal binding regions are rigid in solution even in the absence of an inhibitor, which is confirmed by the relaxation experiments (Fig. 7A). Slight changes for residues in hetNOEs were observed for some residues of eParE in complex with 2. A further H-D exchange experiment revealed that there might be slight conformational changes for residues distant from the binding pocket (Fig. 7). For those residues in structural proximity to the binding site of 2, the inhibitor prevented their exposure to the solvent (Fig. 7C). Based on these H-D exchange experiments, residues that are critical for and affected by 2 binding were revealed, suggesting that this method will also be useful for understanding protein-drug interactions. 19 F NMR experiments are widely used to study ligand-protein binding and dynamics (28,29). The simplicity and high sensitivity (natural abundance of 19 F ϭ 100%) makes the application of such techniques attractive (30). Previous work has seen the application of 19 F NMR of 1 against eParE (17). Similar to 1, chemical shift changes were also observed for 2 when eParE was present, suggesting its binding in solution (Fig. 8). Additionally, a 19 F NMR competition experiment between 1 and 2 allows for ranking of binding potencies between the two compounds and provides a rational prediction of the binding location of inhibitors. In this study, 1 was demonstrated to have both the free and the protein-bound forms when 2 was added into the premixture, indicating competitive binding (Fig. 8B). Furthermore, it was found through this competition experiment that 2 is more potent than 1, corroborating binding results obtained from SPR. As 19 F NMR is as sensitive as 1 H NMR and there is no requirement for isotope labeling (30,31), this method has been proven to be a useful tool for confirming or identifying competitive inhibitors in a drug discovery process. Our results suggested that 1 can serve as a future reference compound for confirming whether a potent GyrB/ParE inhibitor is bound to the ATP site.
In summary, biochemical and biophysical studies were performed to characterize and understand the interactions between eParE and 2. In particular, detailed protein-based NMR studies have confirmed the solution binding mode of 2 against eParE. More importantly, through structural comparison of inhibitors 1 (K D ϭ 902 nM) and 2 (K D ϭ 1.1 nM), it is hypothesized that residues in ␣4, residues close to His-51 and Ala-52, such as Glu-46, Arg-72, and Arg-132, are critical for  19 F NMR of 2 in the absence and presence of different amounts of eParE was collected and processed. B, competition assay using 19 F NMR. 19 F NMR spectra of free 1, eParE-1 complex, and eParE-1/2 complex are shown in green, red, and blue, respectively. Appearance of the 19 F signal corresponding to free 1 suggests that it can be competed out of the ATP binding pocket by inhibitor 2. C, model of using 1 in a competition study.
inhibitor potency against eParE. This in-depth understanding of ligand-ParE interactions at a molecular level will potentially guide and impact future development of topoisomerase inhibitors.

Experimental Procedures
Protein Production and Inhibitor Synthesis-The N-terminal ATP binding domain comprising residues 1-218 of eParE was expressed and purified as described previously (16). Briefly, the pET29b vector harboring the cDNA of eParE and a C-terminal tag (LEHHHHHH) was transformed into E. coli BL21(DE3)competent cells. Recombinant protein was induced at 18°C by adding 1 mM isopropyl 1-thio-␤-D-galactopyranoside overnight when the cell optical density reached 0.8 -1.0. The recombinant eParE was purified using nickel-nitrilotriacetic acid resin and gel filtration chromatography. Protein was concentrated to 0.5-0.8 mM in an NMR buffer that contained 20 mM sodium phosphate (pH 6.5), 150 mM KCl, 2 mM DTT, and 0.5 mM EDTA. Uniformly 13 C/ 15 N, ϳ70% 2 H-labeled eParE was prepared for NOESY data collection. Compounds 1 and 2 ( Fig. 1) were synthesized as described in the literature (11). A stock solution was prepared by dissolving the powder into deuterated DMSO.
Thermal Shift Assay-The thermal shift experiment was carried out using SYPRO Orange dye on a Roche LC480 PCR machine using a similar method as described previously (18,32). A 384-well plate was used in the assay, in which each well contained 10 M eParE with 20 ϫ SYPRO Orange. The assay buffer contained 20 mM sodium phosphate, 150 mM KCl, 2 mM DTT, and 0.5 mM EDTA (pH 7.2). Varying concentrations of 2 from 0 -25 M were mixed with eParE. The samples were subjected to temperature increases from 30 -95°C. GraphPad Prism (GraphPad) was used to generate the melting curve.
IC 50 Assay for eParE-The effect of 2 on ATP hydrolysis was measured, and the IC 50 is defined as the inhibitor concentration that gives 50% of maximum enzymatic activity. All reactions were performed in 30-l volumes in a transparent 384-well plate (17). Mixtures that contained 2 with concentrations varying from 0.39 -200 M were prepared. The assay mixtures also contained 20 mM Tris-HCl (pH 8.0), 8 mM MgCl 2 , 50 mM ammonium acetate, 2.5% (v/v) glycerol, 0.005% (v/v) Brij 35, 0.5 mM EDTA, 5 mM DTT, 2% DMSO, 2 M ParC, 2 M full-length ParE, 160 M ATP, and 0.005 mg/ml salmon sperm DNA (17). The mixtures were first incubated at room temperature for 24 h, and the reaction was quenched by addition of 30 l of a reagent that contained 0.34 mg/ml of malachite green chloride and 0.011 g/ml ammonium molybdate that was prepared in 1 M HCl. The absorbance at 650 nm (A 650 ) was measured after incubation at room temperature for 5 min. The A 650 was plotted against compound concentration (log), and the IC 50 value was obtained.
Surface Plasmon Resonance Measurement-The SPR experiment was carried out to obtain the binding affinity between FIGURE 9. GyrB/ParE-inhibitor interactions. A, structure of the GyrB-C3 complex. The C3 inhibitor is shown as sticks (PDB code 4KFG). Several residues that might be important for improving inhibitor potency are shown as sticks. B, structural model of the eParE-2 complex. The model was from HADDOCK based on CSP caused by 2 binding. Inhibitor 2 is shown as sticks. Glu-46 and Arg-72 might form salt bridges to favor interaction with 2. Residues exhibiting CSP upon inhibitor binding are shown in blue, green, and cyan, respectively. Arg-132 and Asp-133, which showed significant CSPs, are shown in light blue. Met-74 is shown as sticks, and its amide proton exhibited NOEs with water protons. C, 1 H, 15 N HSQC spectra of eParE in the absence and presence of 1 and 2. These three residues exhibited difference when eParE bound to 1 and 2. The peak corresponding to free eParE is shown as a circle. Arrows indicate the direction of peak shifting after inhibitor binding. The assignment of the eParE-1 complex was obtained from a previous report. D, strip plot of Met-74 in the NOESY spectra. The amide proton of Met-74 exhibited NOEs with water protons. For each residue, the left panel is a strip of NOESY of free eParE. The center and right panels are the strips of NOESY and filtered NOESY spectra of the eParE-2 complex, respectively. The peak labeled with a question mark might be a NOE with the inhibitor, which cannot be unambiguously assigned. eParE and 2 on a BIAcore-2000 system (GE Healthcare). CM5 chips were used to analyze protein-2 interactions. The experiment was carried out at 25°C. Purified eParE was first immobilized on a chip. A buffer that contained 10 mM HEPES (pH 7.5), 150 mM NaCl, 3 mM EDTA, and 0.005% v/v surfactant P20 was filtered and degassed. 2 in DMSO was diluted with the buffer before injection. The data were analyzed using software provided with the equipment. The K D value was obtained using a 1:1 steady-state binding model (18).
Ligand-observed 19 F Experiments-As 2 contains only one F atom, 19 F NMR experiments were carried out at room temperature on a Bruker 400 MHz magnet that was equipped with a broadband observe probe. Free inhibitor was prepared in the NMR buffer. For eParE-inhibitor mixtures, different amounts of eParE samples were first lyophilized. The lyophilized sample was then mixed with 2 that was in water instead of buffer because the lyophilized sample contained salts. The spectra were collected and processed using Topspin 2.1. For competition with 1, a protein mixture containing 0.8 mM inhibitor 1 and 0.8 mM eParE was first prepared. A 19 F spectrum was acquired and processed. Different amounts of 2 were then added, followed by 19 F spectrum acquisition and processing. Observation of a peak corresponding to free 1 implies that it can be competed out of the binding pocket by 2.
NMR Experiments-All NMR experiments were collected at 25°C on a Bruker 700-or 600-MHz magnet equipped with a cryoprobe. The backbone assignment of eParE in complex with 2 was obtained based on two-and three-dimensional experiments. Transverse relaxation-optimized spectroscopy (TROSY) (33, 34)-based experiments, including two-dimensional TROSY, three-dimensional HNCACB, HNCOCACB, and HNCO were collected. The collected spectra were processed using NMRPipe (35) and Topspin 2.1. The spectra were analyzed using NMRView (36) and CARA, which was developed in the Wüthrich group. Backbone assignment was conducted based on the three-dimensional experiments and the previous assignment of free eParE (37). The secondary structure was analyzed using TALOSϩ based on the obtained backbone chemical shifts (38).
To map the inhibitor binding site on eParE, 1 H, 15 N TROSY, or HSQC spectra of eParE in the absence and presence of 2 were collected and superimposed. CSP caused by 2 binding were mapped onto the crystal structure of eParE (39). The combined chemical shift changes (⌬␦) were determined using the following equation: ⌬␦ ϭ ((⌬␦ HN ) 2 ϩ (⌬␦ N /5) 2 ) 0.5 , where ⌬␦ HN and ⌬␦ N are the CSPs caused by 2 binding in the amide proton and amide dimensions, respectively (39). The C␣ and C␤ chemical shifts of the complex were also compared with those in free form. To further confirm the model obtained by a docking method, NOEs between protein and the inhibitor was obtained by comparing the NOESY spectra of free eParE and the eParEinhibitor complex and an F1-13 C/ 15 N-filtered F2-15 N-edited NOESY experiment of the complex. A mixing time of 120 ms was used in all NOESY experiments. 15 N-T 1 , T 2 and heteronuclear NOEs (hetNOE) were measured using triple-labeled eParE in the absence and presence of 2. Because of the signal overlap, only hetNOEs were compared for eParE and its complex. T 1 values were obtained from a pseudo-three-dimensional experiment with relaxation delays of 200, 400, 600, 800, 1000, 1200, 1600, and 2000 ms. T 2 values were obtained from a series of two-dimensional 1 H, 15 HSQC spectra from a pseudo-three-dimensional experiment with relaxation delays of 17, 34, 51, 68, 85, 102, 119, 136, and 153 ms. Because of the signal intensities, only the first six data points were used for T 2 determination. The hetNOEs were obtained from spectra collected without and with 3 s presaturation.
The hydrogen-deuterium exchange experiment was conducted using similar methods as described previously (32). Briefly, lyophilized eParE in the absence and presence of an equimolar ratio of 2 were prepared. D 2 O was added to the samples, and a series of 1 H, 15 N HSQC spectra were collected and compared. All experiments were conducted at 298 K, and the spectra were collected and processed as mentioned before.
High Ambiguity-driven Protein-Protein Docking (HADDOCK)-HADDOCK was used to understand protein-inhibitor interactions. The ambiguous interaction restraints were based on residues that exhibited a ⌬␦ of more than 0.3 ppm upon 2 binding. The restraints were generated using the HADDOCK server (40). The eParE structure was obtained from the PDB (PDB code 1SA4). The structure of the inhibitor was generated using the PRODRG server (41). The topology and parameter files of the inhibitor for HADDOCK were generated using HIP-UP (42). The docking was conducted on a local computer using HADDOCK2.1.